Dual-objective optimization for DOA estimation of mixed coherent and noncoherent signals based on DCN
W Wang, Chenxi Mao, Yuehua Chen, Lei Zhang · IET conference proceedings. · 2026
This letter presents a Direction-of-Arrival (DOA) estimation method for mixed coherent and non-coherent signals using a Dual-Objective optimization Deep Convolutional Network (DOO-DCN). The approach leverages DOO-DCN's nonlinear mapping capability to transform mixed signals into ideal noiseless uncorrelated signals through Euclidean distance fitting, while incorporating Toeplitz constraints based on physical properties of covariance matrices from stationary uncorrelated signals received by a Uniform Linear Array (ULA). This dual-objective optimization enables robust detection of mixed and fully coherent signals at low Signal-to-Noise Ratio (SNR) conditions with minimal error using subspace methods. Experimental results demonstrate significant performance advantages over existing methods.